Paranoid how Skynet seems to be all too real nowadays? Artificial intelligence has been rapidly developing into more "self-aware" robots, as Stephen Hawking and Elon Musk already warned the general public about them, specifically how they could end humanity. Now, Google is now taking the necessary steps in stopping any wayward, out of control AI.

Like Us on Facebook

Google's Deepmind director, Laurent Orseau, where this division is also responsible for developing AlphaGo, an AI who defeated Go champion LeesSedol in this ancient, strategy board game, released a scientific paper known as titled "Safely Interruptible Agents". This paper now proposes that Google also establish and incorporate some sort of "kill switch" upon programming any type of AI.

Orseaus and co-author of the study, Stuart Armstrong from Oxford University, reveal new insights about how AI could not adjust and operate in its optimum state in the real world. They also suggest that since AI will be operating under human supervision, and in the event that this AI's strategies and actions result in a negative way that can even become dangerous or harmful to humans, this AI will be considered to be a threat to humans and therefore should be stopped with this so-called "big red button".

In this new paper, the team also presents scenarios where the AI becomes uncooperative and even causing trouble, ensuring that this button will become necessary only if a manual override is not engaged. They also suggest ways to guarantee that this types of AI scenarios will not occur.

An autonomous car equipped with a self-aware AI can lead to disastrous results if it adapts the very human concept of survival known as self-preservation. This can lead to placing passengers and pedestrians at risk, and for events like this, Orseau and Armstrong propose a kill switch.

TagsGoogle Deepmind, AI, Artificial Intelligence, kill switch for AI, google big red button, google kill switch AI